Industry and engineering
Your technical know-how, searchable in seconds and without leaving the company network.
We install an AI in your plant or engineering office that searches drawings, standards, datasheets and incident reports, drafts quotes and reports, and automates processes, without your know-how leaving your network, even with no internet. It protects trade secrets and helps with NIS2. We start with a free assessment at your plant or by video call.
Standards, datasheets, machine manuals, specifications and lessons learned, searchable in natural language with a link to the document and page.
Draft quotes, technical reports and tender responses based on previous projects and the customer's requirements.
Analysis of non-conformities, 8D reports and past complaints to find recurring causes and propose actions.
The technician looks up a machine's manual and breakdown history by its error code and gets the exact page and what was done last time.
Reading purchase orders, delivery notes and certificates, cross-checking them against the ERP and flagging anything that does not match, with the AI running on your server.
Fine-tuning (LoRA) an open model with your terminology, codes and internal formats when search alone is not enough.
Air-gapped deployment for plant networks or R&D departments with no internet connection.
The industrial base of Barcelona and its surroundings (automotive, chemicals, food, machinery, engineering firms and technical offices) depends on knowledge that does not appear in any search engine: tolerances, process recipes, fixes for breakdowns, quotes won and lost. That is exactly what AI can make useful and, at the same time, what should never leave the company network.
Regulation is pushing in the same direction. The Trade Secrets Act only protects what has been protected with reasonable measures, and NIS2 brings digital supplier control to many factories that had never considered it. An AI installed on your own server is not one more supplier: it is a machine you own.
A quality engineer asks what happened the last time a customer rejected a batch because of burrs and gets the three relevant 8D reports, with the action that worked. The person in charge of quotes starts from a draft that already combines the customer’s requirements with two similar projects. In maintenance, the technician looks up an error code and the assistant shows the manual page and that machine’s history. And every morning an agent checks incoming orders and flags any part number that does not match the ERP.
This is not theory. In our Barcelona lab, two clustered NVIDIA DGX Spark units and several RTX GPUs run open models of up to 320 billion parameters, with contexts of up to a million tokens, and agents that work 24 hours a day. Our private gateway has served around 11.15 billion tokens at zero cost per token. We know how much memory each model needs, what fails and when, and how to switch models without interrupting the service.
If the information you want to search is public or there are no trade secrets involved, a cloud tool with a suitable contract may be cheaper. And if the problem is that the documentation does not exist, it has to be written first: AI organises and finds knowledge, but it does not invent it.
AI and GDPR assessment
We review how you use AI today, what data it touches and what obligations you have. You leave with a clear plan: what to run locally, what can go to the cloud and in what order.
Ready-to-use local AI
Your own ChatGPT, inside your office. We choose or reuse the hardware, install the models and set up a private chat for the whole team.
Private RAG search
Ask your documents as you would ask an expert colleague. The AI answers citing the document and page, and respects who can see what.
AI agents
Agents that work with your tools and your rules: they read, draft, update systems and send alerts. With local models and human approval where it matters.
Custom LoRA model
A model that speaks like your business: your jargon, your formats and your tasks. We build the dataset, train the model and measure whether it really improves.
AI maintenance
Your local AI, always in shape: updated models, monitoring, backups and someone who answers when something fails.
Yes. Once installed, it runs inside your network and can work completely isolated (air-gapped). It only needs to go online, if you want it to, to update models during maintenance, and that is done in a controlled way.
It depends on how many people use it and the volume of documentation. For an engineering office of 5 to 30 people, a compact server such as the NVIDIA DGX Spark is usually enough; for many users or agents, two in a cluster, like the one in our Barcelona lab. We help you buy it or we use the hardware you already have.
Not usually at first. Search with citations over your documentation covers most cases; fine-tuning makes sense when there is a vocabulary, format or task so specific to you that a general model cannot handle it.
Yes, if it has an API, an accessible database or exports. The agents read and propose; any write to a critical system goes through human approval.
We install an AI in your plant or engineering office that searches drawings, standards, datasheets and incident reports, drafts quotes and reports, and automates processes, without your know-how leaving your network, even with no internet. It protects trade secrets and helps with NIS2. We start with a free assessment at your plant or by video call.